Human infection by Brucella melitensis: an outbreak attributed to contact with infected goats

Transcription

1 FEMS Immunology and Medical Microbiology 19 (1998) 315^321 Human infection by Brucella melitensis: an outbreak attributed to contact with infected goats Jorge C. Wallach a, Luis E. Samartino b, Adriana Efron c, Pablo C. Baldi d; * a Seccioèn Brucelosis, Hospital F.J. Munìiz, Uspallata 2272, 1282 Buenos Aires, Argentina b Centro de Investigacioèn en Ciencias Veterinarias, Instituto Nacional de Tecnolog èa Agropecuaria (INTA), C.C.77, 1708 Moroèn, Argentina c Laboratorio de Microbiolog èa, Hospital Nacional de Pediatr èa Dr. Juan P. Garrahan, Combate de los Pozos 1881, 1253 Buenos Aires, Argentina d Instituto de Estudios de la Inmunidad Humoral (IDEHU), Jun èn 956 4to. piso, 1113 Buenos Aires, Argentina Received 6 May 1997; revised 20 October 1997; accepted 21 October 1997 Abstract Although several outbreaks of Brucella melitensis infection have been reported among laboratory workers or goat cheese consumers, outbreaks related to rural labour have been rarely studied. An outbreak of human brucellosis among farm workers of Argentina was studied and revealed a close relationship with an epidemic of caprine abortions which occurred shortly before on the same farm. High rates of B. melitensis infection were found among goats. Active brucellosis was diagnosed in 33 subjects (14 with positive blood culture for B. melitensis), while other 27 did not show evidence of illness. While 25 of the brucellosis active patients were rural workers, only 5 of the healthy subjects were engaged in rural labour. Active brucellosis was diagnosed in 91.3% of the subjects in continuous contact with goats and in 32% of those having an occasional contact with the animals. All the 60 subjects denied consumption of goat cheese or milk. As shown here, epidemic human infections by B. melitensis may develop among people frequently in contact with infected goat herds or goat manure. z 1998 Federation of European Microbiological Societies. Published by Elsevier Science B.V. Keywords: Brucella melitensis; Human outbreak; Caprine outbreak 1. Introduction * Corresponding author. Tel.: +54 (1) Fax: +54 (1) Brucellosis, a zoonotic disease caused by several species of Brucella, is a common health problem among people in close contact with cattle or living in rural areas. Human and animal brucellosis are distributed worldwide. The majority of human cases, and the more severe ones, are caused by B. melitensis, although it has a more limited geographical distribution than B. abortus [1]. Transmission of B. melitensis from infected animals to humans may be either direct or indirect. The rst mechanism involves the respiratory, conjunctival and cutaneous routes, and is more important in people in close contact with infected goats or sheep. Conversely, indirect transmission to humans occurs through the consumption of unpasteurized goat milk or cheese by unaware consumers [2]. In contrast to the low risk of transmitting B. abor / 98 / $19.00 ß 1998 Federation of European Microbiological Societies. Published by Elsevier Science B.V. PII S (97)

2 316 J.C. Wallach et al. / FEMS Immunology and Medical Microbiology 19 (1998) 315^321 tus via cow's milk, consumption of sheep or goat milk containing B. melitensis is an important source of human brucellosis worldwide and has caused several outbreaks [3^9]. Human outbreaks related to aerosol transmission in laboratories were also reported [10^12]. Although B. melitensis infection is frequent among people having close contact with infected animals (farmers, veterinarians), mainly in developing countries [13], human outbreaks attributed to contact with infected animals have been rarely reported [14]. An unusually high number of abortions occurred among the goats bred in a farm in the province of Mendoza, Argentina. Shortly afterwards, an outbreak of human brucellosis took place among farm workers. Here we report epidemiological, clinical, serological and bacteriological ndings related to both outbreaks. 2. Materials and methods 2.1. Animals involved in the outbreak A large number of abortions occurred among goats in a rural area of the province of Mendoza, Argentina, where a caprine herd had been recently established. Six months earlier, these goats had had a normal delivery. From the 2200 animals at the time of the outbreak, 400 had been tested by slide agglutination test (Huddleson) when they were admitted to the farm, and all of them were negative by this test. Unfortunately, serological analyzes were not done at admittance for the other 1800 animals. When abortions began, 500 aborting females were separated from the herd and moved to another farm, where no humans lived and no animals were raised. To evaluate the animal outbreak, veterinarians from the National Institute of Agriculture and Farming Technology (INTA-CICV) randomly collected 30 milk and 20 colostrum samples for bacteriological studies. Blood samples for serological studies were also randomly obtained from 276 goats, including 32 males, 156 females with normal delivery, 49 pregnant females and 39 animals that had aborted Human cases In this farm, the breeding area was located 150 metres away from the houses where the workers lived. Fifteen days after the initial goat abortions, one of the goatherds (patient 15) developed an acute illness, characterized by chills, fever, headache, sweats, and rachialgia. Since these symptoms suggested acute brucellosis, blood samples from this worker were obtained for serological tests and blood culture. A positive result (400 IU ml 31 ) was found by the standard tube agglutination and, at the same time, a Brucella spp., later typed as B. melitensis biovar 1, was isolated by blood culture. In view of this, all the persons working and/or living within the farm area underwent a clinical, serological and bacteriological investigation Bacteriological studies Milk and colostrum samples obtained from aborting goats were plated on the Kuzdas-Morse selective medium [15] and incubated in anaerobiosis during 15 days. Brucella spp. was isolated from human blood samples by incubation in a liquid medium containing a casein enzymatic hydrolysate (NZ Amine) and an enzymatic digest of animal tissues (Primatone), both from ICN, Costa Mesa, CA. Strains isolated from animal and human samples were characterized by means of biochemical, serological and phage lysis tests, against reference strains from the Instituto Nacional de Microbiolog èa Carlos G. Malbraèn from Buenos Aires Serological studies Agglutination tests Human sera were assayed by the standard tube agglutination test, with and without 2-mercaptoethanol (2ME-STA and STA, respectively) and the Rose Bengal test [16]. Animal sera were assayed by these tests and also by the bu ered plate antigen (BPA) test [17]. Results of STA and 2ME-STA were expressed as IU ml 31, as indicated by Alton et al. [16], and, for humans, titers higher than 50 IU ml 31 were considered indicative of active brucellosis.

3 J.C. Wallach et al. / FEMS Immunology and Medical Microbiology 19 (1998) 315^ ELISA for detecting antibodies to Brucella LPS Serum reactivity against LPS and LPS-associated proteins was titrated by a capture ELISA, using the anti-lps monoclonal antibody BC68, as described previously [18]. Brie y, polystyrene plates were sensitized with BC68 and, after blocking, were incubated with 5 Wg of Brucella LPS per well. Serial dilutions of patients' sera, starting at 1 in 200, were added and speci c antibodies were detected with an anti-human IgG monoclonal antibody-horseradish peroxidase conjugate (Sigma) or an anti-human IgM polyclonal serum-horseradish peroxidase conjugate (Sigma). The reaction was developed with ortho phenylene diamine and H 2 O 2, and was stopped with 4NH 2 SO 4. The nal color was read at 492 nm. The cut-o value of the assay was calculated as the mean absorbance of 30 sera from healthy controls, assayed at 1 in 200 dilution, plus 3 standard deviations. ELI- SA titers were determined as the last serum dilution giving an absorbance higher than the cut-o value. All the incubations lasted 1 h, except for the color reaction with ortho phenylene diamine ELISA for detecting antibodies to cytoplasmic proteins of Brucella Serum IgG and IgM reactivities against a cytoplasmic protein preparation of Brucella (LPS-free CYT), depleted of LPS by immunoadsorption with BC68 [18], were assayed by indirect ELISA [19]. Polystyrene plates were sensitized with 0.5 Wg/well of LPS-free CYT diluted in PBS. After blocking, sera were assayed at serial dilutions starting at 1 in 100. Addition of the conjugates, development of the reaction and determination of the cut-o value were performed as described above ELISA for detecting antibodies to the 18 kda cytoplasmic protein of Brucella Serum reactivity against an 18 kda cytoplasmic protein of Brucella was titrated by capture ELISA with the monoclonal antibody BI24, as described previously [20]. Puri ed BI24 was adsorbed at 1 Wg/well in PBS onto polystyrene plates. After blocking as described above, 2 Wg/well of the LPSfree CYT antigen were added. Samples were assayed at serial dilutions, starting at 1 in 100. Addition of the conjugates, development of the reaction and determination of the cut-o value were performed as described above Antibiotic therapy Antibiotics, either single or in combination, were used at the following doses: oral doxycycline, 100 mg every 12 h for 42 days; parenteral streptomycin, 1.0 g every 24 h for 21 days; oral rifampin, 300 mg every 12 h during 42 days; oral tetracycline, 500 mg every 6 h for 30 days, parenteral gentamicin, 2.6 mg kg 31 /day for 10 days; oral cipro oxacin, 500 mg every 8 h for 30 days; and oral trimethoprim-sulfamethoxazole (TMS), 160 mg/800 mg every 8 h for 30 days. The initial treatment used was doxycycline + streptomycin or doxycycline+rifampin. The following treatments were subsequently applied for patients with relapse or persistent illness: (a) monotherapy with high doses of TMS, (b) TMS+streptomycin or rifampin, (c) cipro oxacin+gentamicin. 3. Results 3.1. Caprine epizootic B. melitensis biovar 1 was isolated from 17 milk and 11 colostrum samples obtained from goats. From 276 serum samples assayed, 76 (27.5%) showed negative results in all agglutination tests. The remaining 200 sera (72.5%) showed a positive result in STA, with a high proportion of positive reactants found in all groups of goats: 92.3% of the aborting females, 81.2% of the males, 73.5% of the pregnant females and 65.4% of the females that had a normal delivery. In addition, 150 sera were positive by 2ME- STA and 192 were positive by BPA test. Taken into account the high percentage of infection found among tested animals and the risk of undetected infections among those showing negative serological or bacteriological results, veterinarians decided to eliminate all of the 2200 goats Human outbreak Diagnostic aspects After the rst human case was detected, 60 persons at risk underwent clinical and serological exami-

4 318 J.C. Wallach et al. / FEMS Immunology and Medical Microbiology 19 (1998) 315^321 Fig. 1. Time elapsed between caprine outbreak and appearance of human cases of B. melitensis infection. *, Time at which goats were slaughtered. nations for the following 9 months. In 33 subjects, a diagnosis of active brucellosis was established at different times after the animal episode (Fig. 1). This group will be regarded as `patients'. Another 27 asymptomatic subjects, who showed low or negative agglutination titers at initial examination and during the following 6 months, will be regarded as `brucellosis unlikely' subjects. Table 1 summarizes the bacteriological and serological ndings of the active patients at the time of diagnosis. B. melitensis biovar 1 was isolated in 14 (51.8%) of the 27 patients in whom blood cultures could be performed. In the remaining 19 patients, the diagnosis was established on the basis of clinical and serological data. As shown in Table 1, STA titers higher than 50 IU ml 31 were detected in 29 of the active patients, and 9 of them also showed 2ME-STA titers higher than 50 IU ml 31. Anti-LPS antibodies were also detected by capture ELISA in all but one of the active patients. In addition, antibodies against cytoplasmic proteins were detected in the rst sample of all but 5 patients. Eight active patients, with initial low or negative titers of IgG against these proteins, showed rising titers of these antibodies during the follow-up (data not shown). All the 27 brucellosis unlikely subjects were negative by 2ME-STA, but 17 of them had low titers (25 IU ml 31 ) by STA. In addition, low titers of IgM and/or IgG antibodies to Brucella LPS were detected by ELISA in 12 brucellosis unlikely subjects. On the other hand, IgG against LPS-free CYT was negative in 21 of 27 brucellosis unlikely subjects and was positive at low titers (9200) in 6. Similar results were obtained when IgM and IgG antibodies against the 18 kda protein were measured. Objective signs of disease were present in all the 33 active patients and 29 of them also referred symptoms. Splenomegaly was found in 36.3% of the patients, adenitis in 33.3%, hepatomegaly in 30.3%, arthritis in 18.1%, pneumonitis and bronchitis in 9.1%, pharyngitis in 6.0%, and epididymitis, retinitis and icterus in 3.0%. Fever was reported by 89.6% of the symptomatic patients, myalgia by 68.9%, asthenia by 62.0%, rachialgia by 55.2%, headache by 51.7%, sweats by 41.4%, anorexia by 31.0%, arthralgia by 27.6%, cough by 20.6%, and dyspnea and gastrointestinal complaints by 10.3% Epidemiological aspects Patients developed active brucellosis at di erent Table 1 Epidemiological and laboratory features from patients with active brucellosis Occupation Number of patients Contact Blood culture b Number of patients with positive serology by c with goats a C O N + 3 ND STA 2ME anti-lps ELISA Goatherds Other rural labour * 8* Non-rural labour Total anti-prot. ELISA a Contact with goats is classi ed as continuous (C), ocassional (O) or none (N). b Blood culture results are expressed as positive (+), negative (3) or not done (ND). c 2ME-STA and STA: standard tube agglutination with or without 2-mercaptoethanol, respectively. Agglutination was considered positive when the titer was higher than 50 IU ml 31. Anti-LPS ELISA: anti-lps antibodies measured by ELISA, expressed as the sum of IgM and IgG results. Anti-prot.: combined results of IgM and IgG antibodies against LPS-free CYT antigen and the 18 kda cytoplasmic protein. *Serological tests were not done in one patient.

5 J.C. Wallach et al. / FEMS Immunology and Medical Microbiology 19 (1998) 315^ times after the animal episode. As shown in Fig. 1, a rst peak of 16 cases appeared between weeks 6 and 12, while a second peak included 5 cases produced between weeks 16 and 20. The rst peak included 10 of the 16 goatherds that worked on the farm, while the second peak, produced after the goats had been slaughtered (week 14), included only one goatherd. Three cases appeared much later, between weeks 32 and 34. Two of these later cases, however, were masons that had begun to work in the yards later than the other workers. Distribution by sexes among active patients was quite di erent from that found among brucellosis unlikely individuals. From 33 patients with active brucellosis, 32 were male and one was female. In contrast, brucellosis unlikely subjects included 17 women and 10 men. As shown in Table 1, 25 out of the 33 patients were rural workers (including 14 goatherds), while only 5 of the 27 brucellosis unlikely subjects were engaged in rural labour. This di erence suggested that human infection by B. melitensis was related to rural activities, mainly goatbreeding. To con rm this hypothesis, all the 60 persons at risk of exposition were inquired about their contact with possible sources of infection. Fifty-one subjects reported contact with goats, and 23 of them, who either developed daily work with the animals or used to visit the yards every day, were considered to have a `continuous contact'. Twentyeight persons declared sporadic contact with goats and were classi ed as `occasional contact'. Another 9 individuals reported no contact with the animals. Fourteen subjects having continuous contact with goats and 2 individuals having occasional contact with the animals also reported handling goat manure, used as fertilizer. All the 60 inquired persons denied the consumption of goat milk or cheese. As shown in Table 1, 21 (91.3%) of 23 subjects having a continuous contact with goats developed active brucellosis, while only 11 (32%) of 28 having an occasional contact became ill. On the other hand, only 1 (11%) of 9 persons who reported no contact with the animals developed brucellosis. Eighteen of the 21 active patients who declared a continuous contact with goats developed rural activities, while another 3 subjects developed non-rural labour (2 masons, 1 baker). It is worth noting, however, that the 2 masons had been building a new yard, contiguous to the area were goat abortions took place. The baker used to make frequent visits to the yards by his own interest. The 11 active patients having an occasional contact with goats were not involved in goatbreeding but declared to have visited the yards by their own interest. No clear relationship was found between the frequency of contact with goats and the time elapsed, in each case, between the animal outbreak and the onset of human illness. Although 13 of the 16 patients included in the rst peak of cases had a continuous contact with goats, the same was true for 2 of the 5 patients included in the second peak. On the other hand, 3 patients having occasional contact with goats were included in the rst peak of cases and 3 were included in the second peak Antibiotic treatment and clinical outcome Patients started antibiotic treatment as soon as the diagnosis was established. From 33 patients with active brucellosis, 26 were followed-up for 6 to 13 months (mean: 9 months), and other 3 for 4 to 5 months. Unfortunately, 4 patients were lost to follow-up. After an initial treatment, 20 (69%) of the 29 followed patients exhibited clinical recovery, while 7 showed symptoms and signs suggesting persistent illness. Other 2 patients experienced clinical relapse at 6 and 8 months after treatment. After initial antibiotic treatment, follow-up blood cultures were performed in 23 patients, all of them giving negative results. 4. Discussion Human brucellosis still constitutes an important health problem in most developing countries, in which high rates of B. melitensis infection are found among goats [21]. Argentina has one of the highest prevalence rates (higher than 5%) of caprine brucellosis in South America [21]. This situation is facilitated by the tra c of animals without sanitary controls and the extensive trade in goat manure for fertilizer [13]. In this human outbreak, occupational contact with infected goats and/or goat manure seemed to be the principal source of human infection by B. melitensis.

6 320 J.C. Wallach et al. / FEMS Immunology and Medical Microbiology 19 (1998) 315^321 A high number of goats bred on the farm had serological and bacteriological evidence of active infection with B. melitensis and most of the pregnant females aborted. A high discharge of bacteria may be often produced up to 3 months after an infected birth, even by goats undergoing an apparently normal parturition [14]. In the farm where the present outbreak occurred, goat manure was used as soil fertilizer. Brucellae can survive for up to 10 weeks in soil, 7 weeks in faeces and 25 weeks in urine during the cold season [22]. Since the animal outbreak happened during winter and spring months, when the temperature ranged from 5 to 15³C, the bacterium could have conserved its viability in these media for long periods. The 14 goatherds who developed active brucellosis were involved in the collection of goat manure in the yard. Interestingly, 2 beekeepers, who were occasionally engaged in the handling of goat manure, also became infected by B. melitensis. Since faeces were collected in the yard and then shovelled into a truck, workers involved in this activity could became infected by inhalation of contaminated aerosols or by direct contact with faeces and urine. The inhalation of aerosols contaminated with brucellae has long been regarded as an important mean of transmission to humans [10^12,23] and also among goats [2]. Human outbreaks of B. melitensis infection attributed to airborne transmission have occurred in laboratories were this bacterium was being grown [10^12], and airborne-associated outbreaks by B. suis have been also documented among abattoir workers [23]. Abraded skin has been generally accepted as the most probable way of infection following direct contact with infected animals or tissues [3]. Humans can also be infected with Brucella via the conjunctival sac and abattoir workers are strongly encouraged to wear glasses [23]. In this outbreak, the frequency of contact with probable sources of infection, rather than the type of work, appeared to be the main contributory factor for the contagion. Some individuals who were not engaged in goatbreeding, but had close daily contact with the yard where goat abortions took place, also became infected with B. melitensis. All inquired persons denied consumption of goat cheese or milk. Food-borne infection usually results in small outbreaks, involving most family members or persons sharing the same food [6^9]. This kind of transmission seems unlikely in the present outbreak since only the male members of the family, most of which developed rural work, became ill. Almost all goatherds became ill during the rst peak of human cases. The appearance of a second peak after the elimination of goats, could be attributed either to longer incubation times in these patients or to a later contact with bacteria persisting in the environment. The general clinical picture was in accordance with that reported in B. melitensis infections [24]. In a human outbreak in which all patients became infected with B. melitensis by inhalation, the rate of pneumonitis was only 12% [12]. In the present outbreak, there is no direct evidence for an airborne infection, but the presence of pneumonitis in 9% of the patients suggests that, at least in some cases, the infection was acquired by the inhalatory route. Serology proved to be an important diagnostic tool, since it made possible the detection of infected patients even in the absence of a positive blood culture. In this acute human infection, high levels of IgM antibodies against LPS were detected in initial samples, by both agglutination and ELISA tests, in agreement with previous reports [25]. Antibodies against cytoplasmic proteins of Brucella, which constitute appropriate markers of active brucellosis [18,20], were also detected in all but 5 patients. As shown here, breeding of goats lacking sanitary control may result in a serious risk for human health. Although most human cases of B. melitensis infection reported in South American countries are related to ingestion of unpasteurized goat cheese [13], this study demonstrated that contact with infected herds is also a relevant way of transmission in rural areas. Additionally, the use of goat manure as fertilizer appears to be a dangerous practice and local authorities should have to discourage its use. Acknowledgments This research was nanced by grants from the National Research Council (CONICET). P.C.B. is a recipient of a fellowship of CONICET. We are very grateful to Dr. Mabel Regueiro from the Instituto Nacional de Microbiolog èa Carlos G.

7 J.C. Wallach et al. / FEMS Immunology and Medical Microbiology 19 (1998) 315^ Malbraèn for typing studies of bacterial isolates. We are also indebted to Dr. Nidia Lucero, from the same institution. References [1] Corbel, M.J. (1989) Brucellosis: epidemiology and prevalence worldwide. In: Brucellosis: Clinical and Laboratory Aspects (Young, E.J. and Corbel, M.J., Eds.), pp. 25^40. CRC Press Inc., Boca Raton. [2] Crespo Leoèn, F. (1994) Brucelosis ovina y caprina. O ce International des Epizooties, Paris. [3] Nicoletti, P.L. (1989) Relationship between animal and human disease. In: Brucellosis: Clinical and Laboratory Aspects (Young, E.J. and Corbel, M.J., Eds.), pp. 41^51. CRC Press Inc., Boca Raton. [4] Arnow, P.M., Smaron, M. and Ormiste, V. (1984) Brucellosis in a group of travelers to Spain. J. Am. Med. Assoc. 251, 505^ 507. [5] Eckman, M.R. (1975) Brucellosis linked to Mexican cheese. J. Am. Med. Assoc. 232, 636^637. [6] Galbraith, N.S., Ross, M.S., de Mowbray, R.R. and Payne, D.J.H. (1969) Outbreak of Brucella melitensis type 2 infection in London. Brit. Med. J. 1, 612^614. [7] Thapar, M.K. and Young, E.J. (1986) Urban outbreak of goat cheese brucellosis. Pediatr. Infect. Dis. 5, 640^643. [8] Wallach, J.C., Miguel, S.E., Baldi, P.C., Guarnera, E., Goldbaum, F.A. and Fossati, C.A. (1994) Urban outbreak of a Brucella melitensis infection in an Argentine family: clinical and diagnostic aspects. FEMS Immunol. Med. Microbiol. 8, 49^56. [9] Young, E.J. and Suvannoparrat, U. (1975) Brucellosis outbreak attributed to ingestion of unpasteurized goat cheese. Arch. Intern. Med. 135, 240^243. [10] Huddleson, I.F. and Munger, M. (1940) A study of an epidemic due to Brucella melitensis. Am. J. Public Health 30, 944^954. [11] Olle-Goig, J.E. and Canela-Soler, J. (1987) An outbreak of Brucella melitensis infection by airborne transmission among laboratory workers. Am. J. Public Health 77, 335^359. [12] Staszkiewicz, J., Lewis, C.M., Colville, J., Zervos, M. and Band, J. (1991) Outbreak of Brucella melitensis among microbiology laboratory workers in a community hospital. J. Clin. Microbiol. 29, 287^290. [13] Loèpez-Merino, A. (1989) Brucellosis in Latin America. In: Brucellosis: Clinical and Laboratory Aspects (Young, E.J. and Corbel, M.J., Eds.), pp. 151^161. CRC Press Inc., Boca Raton. [14] Alton, G.G. (1991) Brucella melitensis. In: Networking in Brucellosis Research (Frank, J., Ed.), pp. 205^216. United Nations University Press, Tokyo. [15] Kuzdas, C.D. and Morse, E.V. (1953) A selective medium for the isolation of brucellae from contaminated materials. J. Bacteriol. 66, 502^504. [16] Alton, G.G., Jones, L.M., Angus, R.D. and Verger, J.M. (1988) Techniques for the Brucellosis Laboratory. INRA, Paris. [17] Angus, R.D. and Barton, C. (1984) The production and evaluation of a bu ered plate antigen for use in a presuntive test for brucellosis. Dev. Biol. Stand. 56, 349^356. [18] Goldbaum, F.A., Rubbi, C.P. and Fossati, C.A. (1994) Removal of LPS from a Brucella cytoplasmic fraction by a nity chromatography with an anti-lps monoclonal antibody as immunosorbent. J. Med. Microbiol. 40, 174^178. [19] Goldbaum, F.A., Rubbi, C.P., Wallach, J.C., Miguel, S.E., Baldi, P.C. and Fossati, C.A. (1992) Di erentiation between active and inactive human brucellosis by measuring antiprotein humoral immune responses. J. Clin. Microbiol. 30, 604^ 607. [20] Goldbaum, F.A., Leoni, J., Wallach, J.C. and Fossati, C.A. (1993) Characterization of an 18-kilodalton Brucella cytoplasmic protein which appears to be a serological marker of active infection of both human and bovine brucellosis. J. Clin. Microbiol. 31, 2141^2145. [21] Garc èa-carrillo, C. (1990) Animal brucellosis in the Americas. O ce International des Epizooties, Paris. [22] Elberg, S.S. (1981) Guide pour le diagnostic, le traitment et la prophylaxie de la brucellose humaine. World Health Organization, Geneva. [23] Kau man, A.F., Fox, M.D., Boyce, J.M., Anderson, D.C., Potter, M.E., Martone, W.J. and Patton, C.M. (1980) Airborne spread of brucellosis. Ann. N.Y. Acad. Sci. 353, 105^ 115. [24] Young, E.J. (1989) Clinical manifestations of human brucellosis. In: Brucellosis: Clinical and Laboratory Aspects (Young, E.J. and Corbel, M.J., Eds.), pp. 97^126. CRC Press Inc., Boca Raton. [25] Gazapo, E., Gonzalez La Hoz, J., Subiza, J.L., Baquero, N., Gil, J. and de la Concha, E.G. (1989) Changes in IgM and IgG antibody concentrations in brucellosis over time: importance for diagnosis and follow-up. J. Infect. Dis. 159, 219^225.